基于循环神经网络的流量预测算法
首发时间:2019-01-29
摘要:网络流量规模日益增大,依据不同协议的网络流量占比进行检测资源的分配可以有效提高入侵检测设备的性能。准确进行网络流量大小的预测有利于提升入侵检测系统的性能。本文对现有的流量算法进行深入研究,并且对于网络流量时序分布的相关性进行了实验分析,应用协议分析技术,提取TCP数据的数据流特征数据,提出了一种基于RNN的时序的流量预测方案,依据过往流量大小对当前网络不同协议流量进行预测分析。经实验分析证明,会话类数据应用数据流进行预测可以提高预测的准确度,本文的流量预测算法对比AR算法、ARIMA算法,可以有效降低均方根误差,加快收敛速度,提高网络流量预测准确度。
For information in English, please click here
Traffic Prediction algorithm Based on Recurrent Neural
Abstract:The network traffic scale is increasing, and the allocation of detection resources according to the proportion of network traffic of different protocols can effectively improve the performance of the intrusion detection device. Accurate prediction of network traffic size is beneficial to improve the performance of intrusion detection systems. In this paper, the existing traffic algorithm is deeply studied, and the correlation of network traffic timing distribution is analyzed experimentally. The protocol analysis technology is used to extract TCP data stream data and UDP data packet data. A RNN-based timing is proposed. The traffic prediction scheme predicts and analyzes different protocol traffic of the current network accordinTraffic Prediction algorithm Based on RecTraffic Prediction algorithm Based on Recurrent Neuralurrent Neuralg to the past traffic volume. The experimental analysis proves that the application of data stream for prediction of session data can improve the accuracy of prediction. The traffic prediction algorithm of this paper can effectively reduce the root mean square error and improve the accuracy of network traffic prediction by comparing AR algorithm and ARIMA algorithm
Keywords: Information Security Time Series Data Traffic Prediction Recurrent neural network
基金:
引用
No.****
同行评议
勘误表
基于循环神经网络的流量预测算法
评论
全部评论0/1000